AI real-estate platform: an audit, then the fixes in order

The project

An AI platform for real-estate investors, with property pages and valuation and score views. I led its frontend from 2025 to 2026. The app was live and people were using it.

Audit first

I audited the frontend before changing anything. I listed the problems, ranked them by how much they hurt users and fixed them in that order.

What I found and fixed

• Expired sessions failed silently. A user whose login had expired got an empty page, with no message and no way back to sign-in. I sent every request through one shared client with a single handler for expired sessions, and that fixed it on every page.
• A missing setting looked like a backend outage. The app could start without a required setting and then fail on every request. Now it checks its settings at startup and won’t run without them.
• One broken section took down the whole property page. Each section got its own error boundary, so a failure stays inside that section and the rest of the page still works.
• Property pages loaded empty. The pages fetched their data from the browser after they loaded. That was slow, the requests waited on each other, and Google saw an empty page. I moved the data fetching to the server with the Next.js App Router and split the pages into server-rendered sections, so the first response carries the content.

The result

Error reports went down, and more property pages got indexed by Google. All of it was fixed inside the existing app, with no rewrite.

What I check in other apps

When I take over an app, I check these early:

• An expired login sends people back to sign-in.
• The app stops at startup when a setting is missing.
• One broken part can’t take down a whole page.
• The first response carries the content.

Taking over an existing app explains how I work, and fix or rebuild explains how I decide what to keep.

Technologies used: Next.js, TypeScript, React, App Router.